Closed suminexrepo closed 7 months ago
Managed to try llamafile and it seems like it thinks I have an AMD gpu?
initializing gpu module... extracting /zip/llama.cpp/ggml.h to C:\Users\Computer/.llamafile/ggml.h extracting /zip/llamafile/compcap.cu to C:\Users\Computer/.llamafile/compcap.cu extracting /zip/llamafile/llamafile.h to C:\Users\Computer/.llamafile/llamafile.h extracting /zip/llamafile/tinyblas.h to C:\Users\Computer/.llamafile/tinyblas.h extracting /zip/llamafile/tinyblas.cu to C:\Users\Computer/.llamafile/tinyblas.cu extracting /zip/llama.cpp/ggml-impl.h to C:\Users\Computer/.llamafile/ggml-impl.h extracting /zip/llama.cpp/ggml-cuda.h to C:\Users\Computer/.llamafile/ggml-cuda.h extracting /zip/llama.cpp/ggml-alloc.h to C:\Users\Computer/.llamafile/ggml-alloc.h extracting /zip/llama.cpp/ggml-backend.h to C:\Users\Computer/.llamafile/ggml-backend.h extracting /zip/llama.cpp/ggml-backend-impl.h to C:\Users\Computer/.llamafile/ggml-backend-impl.h extracting /zip/llama.cpp/ggml-cuda.cu to C:\Users\Computer/.llamafile/ggml-cuda.cu note: won't compile AMD GPU support because $HIP_PATH/bin/clang++ is missing extracting /zip/ggml-rocm.dll to C:\Users\Computer/.llamafile/ggml-rocm.dll dynamically linking C:\Users\Computer/.llamafile/ggml-rocm.dll GPU support successfully linked and loaded CUDA error: no ROCm-capable device is detected current device: -1, in function ggml_init_cublas at ggml-cuda.cu:6979 hipDeviceSynchronize() GGML_ASSERT: ggml-cuda.cu:386: !"CUDA error"
I used to have an AMD card in this machine. Any ideas on how to fix this?
I managed to solve this issue by downloading the amd cleanup utility and letting it run while in safe mode.
Describe the bug
It just fails and uses the CPU instead, not sure what the issue is. I use oobabooga separately and can load models with gpu offloading via llama.cpp, so I don't know. Any suggestions?
Steps to reproduce
No response
LLMUnity version
81d38bb6eb7b09aeb483c924c8153ecdfd2d25be
Operating System
Windows